This reverts commit 5f4a945182.
The cure is worse than the original disease.
The main installation cursor often holds exclusive locks (due to DDL
changes) on vital tables such as res_users. As a result, using
another cursor to perform changes while the main cursor is waiting
is extremely deadlock-prone. And these deadlocks can't be detected by
PostgreSQL as they mix Python-SQL locking, which leads to deadlocked
HTTP workers.
Related to:
- opw-1916918
- #29528
- Create a new DB with `base_automation`, without demo data
- Switch to developer mode
- Go to Settings, then 'Load demo data', validate
A traceback occurs.
An error since the demo data in `base_automation_demo.xml` are pointing
to `model_base_automation_lead_test`, which is a test model. It is
therefore not created whithout demo data. The error is:
```
bad query: b'RELEASE SAVEPOINT "5159....."'
```
However, that's not the real crash. Indeed, in case the installation of
demo data crashes, `demo_failure_todo` is supposed to handle the
situation like a boss. But in this case, the DB cursor seems unusable.
Therefore, the following crashes like a gros caca:
``` python
todo = env.ref('base.demo_failure_todo', raise_if_not_found=False)
```
Solution: use a new cursor.
opw-1916918
closesodoo/odoo#29528
This is necessary to get --dev=xml working on windows, because the xml
import uses normcase in file_open and the case normalized filename is
then used to find the addon path (which was not found before on windows)
With this fix arch_fs is now correctly filled in windows and --dev=xml
works.
closesodoo/odoo#28331
After a module upgrade, validate the architecture of the module's views that
are impacted by updates, but have not been checked yet. Before this patch,
views were checked twice on average.
This patch speeds up the installation of modules by about 15% without demo
data. With demo data, the speedup is around 10%.
The registry loading system should not alter databases unless it is
asked to do so, by a module installation or update instruction.
This property should hold true as well for database bootstrap, and this
is what this commit changes..
In order to avoid any behavior change for command-line users, an
implicit `-i base` is assumed when starting the server from the
command-line with `-d <db>`, causing the db boostrap to happen if the
database did not exist yet.
Task 1856935
Currently, if installing a module's demo data fails (because the
module was uninstalled and some were left over and can't be
reinstalled, or because they're broken, or…) the installation of the
module fails and further installations are skipped (?).
Since demo data are non-essential (though required to run tests),
rather than fail everything if they fail just roll them back and
notify the user.
This change generates a warning log *and* shows a notification popup
to the end user.
Try to remove cr.commit (and rollback) from module and db install:
* put a savepoint around test data loading
* remove a bunch of commits sprinkled throughout
* remove rollback on data loading failure (assuming it bubbles up, the
entire module's installation should be rolled back)
* add commit right before the tests are run, so they can run isolated
and still see whatever was done when installing their module
* convert a few explicit closing to context managers
Before this commit:
* Module A defines a field X of model M
* Module B inherits from model M without touching field X
* Module C inherits from model M and extends field X by giving an INDEX
/ NOT NULL constraint.
* Module B and C depend from Module A, but not each other
If all three modules are installed and Module B is updated, the INDEX /
NOT NULL constraint could be dropped.
This happens because Module B can be loaded before Module C is loaded,
if that's the case, then after the upgrade of Module B, during the
schema checking, we verify that the field object we have and the field
on the DB are the same, since Module B doesn't introduce the index then
this check is false and we drop the index. When we get to loading Module
C, we do not do any schema checking because the module is not marked as
`to upgrade`, therefore the index is lost forever.
To solve this, we re-init the models that belong to the set of the intersection
between upgraded and modified models and loaded and modified models.
Fixes#24958
Before this commit:
* Module A defines a field X of model M
* Module B inherits from model M without touching field X
* Module C inherits from model M and extends field X by giving an INDEX
/ NOT NULL constraint.
* Module B and C depend from Module A, but not each other
If all three modules are installed and Module B is updated, the INDEX /
NOT NULL constraint could be dropped.
This happens because Module B can be loaded before Module C is loaded,
if that's the case, then after the upgrade of Module B, during the
schema checking, we verify that the field object we have and the field
on the DB are the same, since Module B doesn't introduce the index then
this check is false and we drop the index. When we get to loading Module
C, we do not do any schema checking because the module is not marked as
`to upgrade`, therefore the index is lost forever.
To solve this, we re-init the models that belong to the set of the intersection
between upgraded and modified models and loaded and modified models.
Fixes#24958
Before this patch the registry and cache signaling was only activated
for PreforkServer. In case Odoo was deployed in a multi process/multi
threaded architecture the signaling was not ensured, causing registry
de-synchronisation amongst threaded servers.
Backported from fd4939dc78
Christophe Simonis